HAWTHORNE, Calif. — Hours after SpaceX ended its first day of public trading at a $2.1 trillion valuation, Elon Musk posted a brief message on X that sent a clear signal about where he intends to take the company next: "Looking forward to taking our exciting partnership with Nvidia to the next-level."
The post came in direct response to Nvidia CEO Jensen Huang's public congratulations to SpaceX for its historic Nasdaq debut — and it landed the same day SpaceX officially unveiled the technical specifications for its AI1 satellite, the company's first dedicated orbital compute platform. The combination of announcements confirms that SpaceX is moving quickly from launch services and internet connectivity into AI infrastructure at a planetary scale. The orbital AI data center strategy has been building for months, with Musk describing space as vast enough to accommodate a million AI satellites.
AI1: SpaceX's Orbital Compute Satellite
The AI1 satellite carries a peak compute payload of 150 kilowatts, with an average sustained compute of 120 kilowatts and a power-to-weight ratio of 70 kilowatts per ton. The platform features a 70-meter wingspan for solar array deployment, liquid radiators for heat dissipation at those power levels, meteorite shielding, and a centralized compute module designed to accept interchangeable compute hardware from multiple providers.
That last point — provider-interchangeable compute — is significant. It means SpaceX's AI1 satellites are not locked to any single chip architecture, giving the company flexibility to populate them with Nvidia hardware, custom silicon, or future processors depending on cost and performance requirements at the time of launch. The satellites will be manufactured at SpaceX's Gigasat facility in Texas.
Terafab: 100 Million Square Feet, 1 Terawatt Per Year
Alongside the AI1 reveal, Musk provided new scale details on Terafab, the joint advanced chip fabrication facility SpaceX is building with Tesla and Intel near Gigafactory Texas in Austin. Musk described Terafab as targeting roughly 100 million square feet of floor space — approximately 10 times larger than Giga Texas — with an annual output goal of 1 terawatt of compute capacity. For comparison, total annual US electricity consumption currently runs at about 0.5 terawatts, illustrating the scale of compute Terafab is designed to produce.
Nvidia's role in both the AI1 satellite and Terafab programs is already substantial. Google currently pays SpaceX $920 million per month for compute capacity equivalent to 110,000 Nvidia GPUs. Anthropic holds a separate agreement for 220,000 Nvidia GPUs at $1.25 billion per month. Both deals run through SpaceX's infrastructure and depend on Nvidia hardware at the core.
A Deepened Partnership Takes Shape
Musk's tweet is not a formal announcement, but the context makes the direction clear. SpaceX has more Nvidia GPU demand from enterprise customers than almost any other infrastructure operator, it is building satellite compute platforms specifically designed to run AI workloads at low Earth orbit, and it is co-developing a chip fab at a scale with no precedent in the private sector.
A next-level Nvidia partnership in this context could mean co-design of space-rated AI chips, preferential GPU allocation for the growing orbital data center pipeline, or collaborative development of hardware that meets the extreme power and thermal constraints of the AI1 platform. The full reporting on Musk's announcement and the AI1 satellite reveal is available from Wccftech.
SpaceX's transition from launch company to AI infrastructure provider is moving faster than most observers anticipated, and Nvidia appears to be the hardware partner for the journey.
